One of the main clinical problems in Barrett's esophagus (BE), like most pre-malignant neoplasms, is predicting which patients are likely to progress to cancer (i.e., risk stratification). Once we identify high risk patients, we can justify interventions to prevent development of cancer. However, neoplasms progress to cancer through a fundamentally stochastic process of somatic evolution leading to clonal heterogeneity in both space and time. Recent cancer genome sequencing results show that many different combinations of mutations can generate a cancer. All of this makes it difficult to develop reliable and accurate panels of mutations for risk stratification. As leaders in the evolutionary biology of cancer and agent-based modeling of neoplastic progression, with access to tissue samples and data from the world's best prospective cohorts of BE, we are in a unique position to study the evolutionary dynamics of neoplastic progression in BE and translate our results to the clinic. Our preliminary results show that by measuring the parameters of somatic evolution, rather than its products, we can develop robust biomarkers for risk stratification that may be universally applicable to most, if not all neoplasms. Specifically, we have shown that measures of genetic diversity, which is the fuel of natural selection, predict progression in BE. We also demonstrated that we can measure the somatic mutation rate, in vivo, and that non-steroidal anti-inflammatory drugs, which appear to prevent cancer in BE, were associated with an order of magnitude decrease in that mutation rate in most patients. This novel initial view of the dynamics of progression revealed 3 surprises that are important for risk stratification and cancer prevention in BE: (1) The mutation rate at the biopsy level is apparently very low: ~1 chromosomal lesion per year, per cell lineage. Aim 1 will test if the mutation rate within crypts is higher than can b observed in biopsies, both in patients that progressed to cancer, and those that did not. (2) The vast majority of lesions occur prior to the first endoscopy, which is often concomitant with the start of powerful acid suppressive medications such as proton-pump inhibitors (PPIs). Aim 2 will use an observational cross-over study design to test the effects of PPIs on mutation rate at the biopsy and crypt levels. This would be the first time that PPIs would be tested for a cancer prevention mechanism. (3) Current theory posits that neoplastic progression proceeds through a series of (e.g., ~20) clonal expansions. Yet, we only observed one such expansion in 156 patient years of surveillance based on biopsy sampling. Aim 3 will develop methods to detect clonal expansions in the much finer resolution available in single cells from cytology brushings of BE from two large, independent cohorts, and test if the detection of a clonal expansion predicts progression. We will also use these cohorts as an independent validation study to test if cell level genetic diversity predicts progression. For each aim, we use computational models to generate the expected data under alternative hypotheses and to identify the most sensitive aspects of the somatic evolutionary process for cancer prevention.